Isolated Bangla word recognition and speaker detection by semantic modular time delay neural network (MTDNN)
Md. Yasin Ali Khan, Sazzad Hossain, Mohammed Moshiul Hoque · 2015
Speaker recognition is the identification of a person from characteristics of his/her voices and speech recognition concerns the recognizing of what is being said by the speaker. This paper presents a framework to recognize the isolated Bangla words and the corresponding speaker by proposing a semantic modular time delay neural network (MTDNN). Underlying acoustic fuzziness of human utterance and fluctuations of data due to environmental disturbance are managed by well-known Fuzzy C Means clustering technique. We have used MFCC features to recognize Bangla words and speaker detection. Experimental result with different individuals show that the proposed framework is functioning quite satisfactory with average accuracy of 82.66%.